526454_en-US

liliconcom

Stability Analysis Calculation Question

A customer has a question about the Stability Platform in JMP. In the cross-check of the fit model and the stability platform, he noticed that different expiry dates are calculated, especially for the different intercept and different slope models. This is also briefly described in the video (Stability Analysis - JMP for Pharmacy - Dr Byron). Apparently the stability platform uses unpooled MSE for the batches and fit model uses the pooled MSE. So far so good. However, the FDA recommends in Q 1 E Evaluation of Stability Data (europa.eu) (B.2.2.1): The retest periods or shelf lives for individual batches in the stability study can be estimated by applying the approach described in Section B. 1 using individual intercepts and individual slopes and the pooled mean square error calculated from all batches.

So his question would be, why does the stability platform use a different calculation method here than specified in the guideline? He didn't really find anything about this in the community, only that JMP refers to a calculation method according to Chow (see also Solved: Question about Stability Analysis (ANCOVA) and the pooled MSE - JMP User Community in the question) and that's the difference is apparently known about the guideline (see also Assumptions of JMP Stability Test - JMP User Community , last post).

Re: Stability Analysis Calculation Question

The procedure that we developed up to JMP 16 was based upon the reference that you mentioned by Chow 2007, more specifically, "STAB", a SAS macro that is provided by the book. The original authors of the macro were FDA researchers. The macro was developed according to the first edition of the guideline, published in 1987.

It was known to us that there are discrepancies on how to interpret guidelines, and we were aware of guidelines change over time.

It was a decision to implement STAB procedure, due to various reasons, especially that we knew the STAB macro was used in the industry at the moment of implementation.

Since then, we have heard requests to implement the pooled MSE approach that is described in the most recent guideline.

The usual implication of using STAB's unpooled MSE, versus the pooled MSE for the nonpoolable batches is that the expiration date will likely be earlier by using the STAB approach, if batches are not poolable.

In JMP 17, we implemented that option, see the next screenshot:

One should check that highlighted box, if a pooled MSE is desired. At this moment, I implemented it as an option, so that existing and new behaviors are both available.

Re: Stability Analysis Calculation Question

In v16 and earlier, the pooled MSE is not used in the individual slopes model.  In v17, the using the pooled MSE will be available as a platform preference, as this has been a frequent request.   You can see this in JMP 17 EA9.  

It is also a scriptable setting in the saved platform script.  

 

Degradation(
 Y( :"Concentration (mg/Kg)"n ),
 Time( :Time ),
 Label( :Batch Number ),
 Connect Data Markers( 0 ),
 Show Fitted Lines( 1 ),
 Show Spec Limits( 1 ),
 Show Median Curves( 0 ),
 Show Legend( 1 ),
 No Tab List( 0 ),
 Set Lower Spec Limit( 99 ),
 Use Pooled MSE for Nonpoolable Model( 1 ),
 Set Censoring Time( . ),
 Show Residual Plot( 1 ),
 Show Inverse Prediction Plot( 1 ),
 Show Curve Interval( 1 ),
 Inverse Prediction Interval( Confidence Interval ),
 Inverse Prediction Alpha( 0.05 ),
 Inverse Prediction Side( Lower One Sided ),
 Test Stability
);